The nature of antimony-enriched surface layer of Fe–Sb mixed oxides

7
The nature of antimony-enriched surface layer of Fe–Sb mixed oxides Yan Huang a, * , Patricio Ruiz b a College of Chemistry and Chemical Engineering, Nanjing University of Technology, Xin-Mo-Fan-Road 5, Nanjing 210009, PR China b Unite de Catalyse et Chimie des Mate ´riaux Divise ´s, Universite ´ Catholique de Louvain, B-1348 Louvain-la-Neuve, Belgium Received 31 July 2005; received in revised form 22 September 2005; accepted 22 September 2005 Available online 26 October 2005 Abstract Antimony segregation is a common feature in Fe–Sb mixed oxides, which have been widely applied as catalysts in selective oxidation and ammoxidation reactions. This paper attempts to shed a light on the cause of such a common feature and on the nature of the antimony-enriched surface layer over FeSbO 4 by means of XPS surface analysis. Single-phase FeSbO 4 samples prepared by different methods were studied, and the antimony in their surface layer is a mixture of both Sb 5+ and Sb 3+ rather than single Sb 5+ . Their surface composition is close to FeSb 2 O 6 , which could be described as (FeSbO 4 )(Sb 2 O 4 ) d , d = 0.5, and it is not ‘‘Fe(II)Sb(V) 2 O 6 ’’ as suggested in literature. Fe–Sb mixed oxides with Sb/Fe > 1 (mol/mol) are mixtures of FeSbO 4 and Sb 2 O 4 , and the surface of FeSbO 4 grains would be a layer of (FeSbO 4 )(Sb 2 O 4 ) d , d 0.5. Fe–Sb mixed oxides with Sb/Fe < 1 are mixtures of FeSbO 4 and Fe 2 O 3 , and the surface of FeSbO 4 grains would be a layer of (FeSbO 4 )(Sb 2 O 4 ) d , d 0.5, but the remaining Fe 2 O 3 would be encapsulated by a layer of FeSbO 4 . # 2005 Elsevier B.V. All rights reserved. Keywords: Surface segregation; X-ray photoelectron spectroscopy; FeSbO 4 ; FeSb 2 O 6 ; Sb 2 O 4 ; Fe 2 O 3 1. Introduction Fe–Sb mixed oxides with Sb/Fe > 1 (mol/mol), which are composed of FeSbO 4 and Sb 2 O 4 , have been extensively applied as catalysts for selective oxidation and/or ammoxidation of propylene, propane, butylenes, methanol, etc. [1–17]. FeSbO 4 is well accepted to be the active component, but this does not mean FeSbO 4 itself is the active species because the surface, rather than the bulk, of FeSbO 4 grains is the deciding factor on catalysis. Even in single-phase FeSbO 4 samples, the surface of FeSbO 4 grains is not simply the FeSbO 4 species but an antimony-enriched layer [8–13]. In addition, conventional Fe– Sb catalysts contain a big amount of excess Sb 2 O 4 , it will enhance antimony enrichment on FeSbO 4 surface. Therefore, it is the antimony-enriched layer that possesses the active species or active sites. Most of the investigations on Fe–Sb catalysts have been associated with surface studies. However, the nature of FeSbO 4 surface layer is still controversial so far. In literature, this antimony-enriched layer over FeSbO 4 grains has been interpreted roughly as: (1) a layer of FeSb x O y (x > 1) species, e.g. Fe(II)Sb(V) 2 O 6 [7–10]; (2) a layer of solid solution of Sb 2 O 4 –FeSbO 4 [13,14]; (3) a thin ‘‘skin’’ [15] or epitaxy [16] of Sb 2 O 4 over FeSbO 4 . Numerous experiments have clearly proved that the best catalytic performances can be reached only when Sb/Fe > 1, no matter how Fe–Sb catalysts are prepared. Hence, this paper is not to report more experiment to that family but attempts to shed a light on the cause of such a common feature and on the nature of the antimony-enriched surface layer via XPS surface analysis. 2. Experimental 2.1. Preparation 2.1.1. Co-precipitation SbCl 3 (p.a. Fluka) was dissolved in 6 mol/l solution of HCl, a little heating was done if precipitate appeared (due to the hydrolysis of SbCl 3 to SbOCl), then stoichiometric Fe(NO 3 ) 3 9H 2 O (p.a. Fluka) was dissolved into the above SbCl 3 solution, and brown gas (NO 2 ) was liberated at this stage. The resulting solution was added drop wise into 1 mol/l www.elsevier.com/locate/apsusc Applied Surface Science 252 (2006) 7849–7855 * Corresponding author. Tel.: +86 25 83587503; fax: +86 25 83365813. E-mail address: [email protected] (Y. Huang). 0169-4332/$ – see front matter # 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.apsusc.2005.09.055

Transcript of The nature of antimony-enriched surface layer of Fe–Sb mixed oxides

The nature of antimony-enriched surface layer

of Fe–Sb mixed oxides

Yan Huang a,*, Patricio Ruiz b

a College of Chemistry and Chemical Engineering, Nanjing University of Technology,

Xin-Mo-Fan-Road 5, Nanjing 210009, PR Chinab Unite de Catalyse et Chimie des Materiaux Divises, Universite Catholique de Louvain,

B-1348 Louvain-la-Neuve, Belgium

Received 31 July 2005; received in revised form 22 September 2005; accepted 22 September 2005

Available online 26 October 2005

www.elsevier.com/locate/apsusc

Applied Surface Science 252 (2006) 7849–7855

Abstract

Antimony segregation is a common feature in Fe–Sb mixed oxides, which have been widely applied as catalysts in selective oxidation and

ammoxidation reactions. This paper attempts to shed a light on the cause of such a common feature and on the nature of the antimony-enriched

surface layer over FeSbO4 by means of XPS surface analysis. Single-phase FeSbO4 samples prepared by different methods were studied, and the

antimony in their surface layer is a mixture of both Sb5+ and Sb3+ rather than single Sb5+. Their surface composition is close to FeSb2O6, which

could be described as (FeSbO4)(Sb2O4)d, d = 0.5, and it is not ‘‘Fe(II)Sb(V)2O6’’ as suggested in literature. Fe–Sb mixed oxides with Sb/Fe > 1

(mol/mol) are mixtures of FeSbO4 and Sb2O4, and the surface of FeSbO4 grains would be a layer of (FeSbO4)(Sb2O4)d, d � 0.5. Fe–Sb mixed

oxides with Sb/Fe < 1 are mixtures of FeSbO4 and Fe2O3, and the surface of FeSbO4 grains would be a layer of (FeSbO4)(Sb2O4)d, d � 0.5, but the

remaining Fe2O3 would be encapsulated by a layer of FeSbO4.

# 2005 Elsevier B.V. All rights reserved.

Keywords: Surface segregation; X-ray photoelectron spectroscopy; FeSbO4; FeSb2O6; Sb2O4; Fe2O3

1. Introduction

Fe–Sb mixed oxides with Sb/Fe > 1 (mol/mol), which are

composed of FeSbO4 and Sb2O4, have been extensively applied

as catalysts for selective oxidation and/or ammoxidation of

propylene, propane, butylenes, methanol, etc. [1–17]. FeSbO4

is well accepted to be the active component, but this does not

mean FeSbO4 itself is the active species because the surface,

rather than the bulk, of FeSbO4 grains is the deciding factor on

catalysis. Even in single-phase FeSbO4 samples, the surface of

FeSbO4 grains is not simply the FeSbO4 species but an

antimony-enriched layer [8–13]. In addition, conventional Fe–

Sb catalysts contain a big amount of excess Sb2O4, it will

enhance antimony enrichment on FeSbO4 surface. Therefore, it

is the antimony-enriched layer that possesses the active species

or active sites. Most of the investigations on Fe–Sb catalysts

have been associated with surface studies. However, the nature

of FeSbO4 surface layer is still controversial so far. In literature,

this antimony-enriched layer over FeSbO4 grains has been

* Corresponding author. Tel.: +86 25 83587503; fax: +86 25 83365813.

E-mail address: [email protected] (Y. Huang).

0169-4332/$ – see front matter # 2005 Elsevier B.V. All rights reserved.

doi:10.1016/j.apsusc.2005.09.055

interpreted roughly as: (1) a layer of FeSbxOy (x > 1) species,

e.g. Fe(II)Sb(V)2O6 [7–10]; (2) a layer of solid solution of

Sb2O4–FeSbO4 [13,14]; (3) a thin ‘‘skin’’ [15] or epitaxy [16]

of Sb2O4 over FeSbO4.

Numerous experiments have clearly proved that the best

catalytic performances can be reached only when Sb/Fe > 1, no

matter how Fe–Sb catalysts are prepared. Hence, this paper is not

to report more experiment to that family but attempts to shed a

light on the cause of such a common feature and on the nature of

the antimony-enriched surface layer via XPS surface analysis.

2. Experimental

2.1. Preparation

2.1.1. Co-precipitation

SbCl3 (p.a. Fluka) was dissolved in 6 mol/l solution of

HCl, a little heating was done if precipitate appeared (due

to the hydrolysis of SbCl3 to SbOCl), then stoichiometric

Fe(NO3)3�9H2O (p.a. Fluka) was dissolved into the above

SbCl3 solution, and brown gas (NO2) was liberated at this stage.

The resulting solution was added drop wise into 1 mol/l

Y. Huang, P. Ruiz / Applied Surface Science 252 (2006) 7849–78557850

NH3�H2O solution under vigorous stirring, and NH3�H2O

(32%) was added manually to keep PH value to ca. 8. The

resulting precipitate was washed with distilled water and

subsequently with acetone. After drying in ambient air, the

precursor was calcined 1.5 8C/min in the air at 800 8C for 5 h.

The resulting samples are noted as CP-Fe2Sb, CP-FeSb and CP-

FeSb2, their corresponding atomic ratio of Sb/Fe is 0.5, 1 and 2,

respectively. For reference, a sample of antimony oxide and a

sample of iron oxide were obtained under similar conditions but

without adding Fe(NO3)3�9H2O or SbCl3. They are noted as

CP-Fe and CP-Sb.

2.1.2. Sol–gel method

Anhydrous citric acid 0.06 mol was dissolved in 10 ml

distilled water, 0.04 mol SbCl3 and 0.02 mol Fe(NO3)3�9H2O

were then dissolved into the above solution under stirring, brown

gas (NO2) was liberated. The solution was evaporated by boiling

until a gel then a solid was obtained. During this procedure, Cl�

can be removed through volatile HCl. The solid was ground and

heated in the air at 500 8C for 4 h to decompose the organic com-

ponent. At last, the precursor was calcined (1.5 8C/min) in the air

at 800 8C for 5 h. The resulting sample is noted as SG-FeSb.

2.1.3. Mechanical method

Stoichiometric hematite Fe2O3 (p.a. Fluka) and senarmon-

tite Sb2O3 (p.a. Fluka) were mixed mechanically, some acetone

was added to help grinding. After drying at 100 8C overnight,

the mixtures with Sb/Fe = 0.5, 1 and 2 (mol/mol) were heated at

a rate of 1.5 8C/min and calcined at 1000 8C for 5 h, the

resulting samples were noted as MC-Fe2Sb, MC-FeSb and MC-

FeSb2. For reference, Fe2O3 and Sb2O3 were calcined at the

same condition and noted as MC-Fe and MC-Sb.

2.2 Characterization

The specific surface areas were measured with Micro-

meritics ASAP 2000 equipment. XRD analysis was conducted

on Seipert XRD 3000 Diffractometer with Cu Ka radiation.

SEM characterizations were carried out on Philips XL40

Scanning electron microscope. FT-Raman spectra were

recorded with a Bruker RFS100 spectrometer coupled with a

diode-pumped germanium solid detector, cooled in liquid

nitrogen, using a Nd:YAG laser as exciting source, the laser

power was set to 100 mW. XPS measurements were performed

Table 1

Results of XRD, surface area and XPS

Sample Phase Area (m2/g) Atomic ratio

O/Sb Sb/F

CP-FeSb2 Sb2O4 + FeSbO4 28 2.5 4.0

MC-FeSb2 Sb2O4 + FeSbO4 1.0 2.4 5.9

CP-FeSb FeSbO4 38 3.1 1.8

MC-FeSb FeSbO4 1.2 3.2 2.1

SG-FeSb FeSbO4 17 3.2 2.0

CP-Fe2Sb Fe2O3 + FeSbO4 41 2.6 1.1

MC-Fe2Sb Fe2O3 + FeSbO4 1.3 3.7 1.6

a Full width at half maximum of Sb 3d3/2 peak.

with an X-ray photoelectron spectrometer SSX-100 model 206

from FISONS [18,19]. X-rays produced by a monochromatized

Al anode (Al Ka = 1486.6 eV) were focused on around

1.4 mm2 spot. The flood gun energy was set to 10 eV with a

fine meshed nickel grid placed 3 mm above the sample surface.

The spectra of survey spectrum, C 1s, Sb 3d (together with

overlapped O 1s) and Fe 2p were recorded subsequently, and

finally C 1s spectrum was recorded again to check the stability

of charge compensation. The binding energies were referenced

to 1s peak of adventitious carbon bound only to carbon and

hydrogen at 284.8 eV. After a Shirley-type background

subtraction in order to minimize the background effect, the

spectra were decomposed with a Gaussian/Lorentzian percent

function of 85/15%. The XPS intensities were calculated

using atomic sensitivity factors provided by the spectrometer

manufacturer. Peak areas of Fe 2p (including Fe 2p1/2, Fe 2p3/2

and their shake-up peaks), Sb 3d3/2 and O 1s bands were used to

quantify Fe, Sb and O. Although Sb 3d5/2 peak overlaps O 1s,

peak area of O 1s can be calculated taking into account that,

theoretically, the distance between Sb 3d5/2 and Sb 3d3/2 peaks

is 9.34 eV and the peak area of Sb 3d5/2 is 1.5 times that of

Sb 3d3/2.

3. Results

The results of XRD phase analysis are listed in Table 1.

CP-Fe2Sb and MC-Fe2Sb are mixtures of FeSbO4 and Fe2O3,

CP-FeSb2 and MC-FeSb2 are mixtures of FeSbO4 and Sb2O4.

Samples of CP-FeSb, SG-FeSb and MC-FeSb were found to be

single-phase FeSbO4, their XRD patterns are shown in Fig. 1. It

should be noted that SbCl3/Fe(NO3)3 = 2 (mol/mol) has been

applied to prepare SG-FeSb, that is, half of the antimony was

lost during preparation. Our preliminary tests indicate that the

starting recipe of SbCl3/Fe(NO3)3 = 1 (mol/mol) would finally

result in a mixture of FeSbO4 and Fe2O3 after sol–gel procedures.

This was confirmed by Carrazan et al. [3], who used similar sol–

gel technique but with a recipe of SbCl5/Fe(NO3)3 = 1 (mol/mol)

to prepare FeSbO4, and impurity Fe2O3 phase was found in

their FeSbO4 sample. During this preparation experiment,

SbCl3 was firstly oxidized into SbCl5 on contacting with NO3�:

Sb3þ þ 2NO3� þ 4Hþ ! Sb5þ þ 2H2O þ 2NO2 " :

SbCl5 is volatile with low boiling point, and the excess SbCl5was vaporized during subsequent steps.

Binding energy (eV) FWHMSba(eV)

e (Rs) Sb 3d3/2 Fe 2p3/2 O 1s

540.1 711.2 530.4 1.88

540.0 711.2 530.4 2.15

539.9 711.1 530.3 1.74

539.8 711.0 530.4 1.92

539.9 711.1 530.2 1.81

540.0 711.0 530.3 1.59

539.9 710.9 530.2 1.74

Y. Huang, P. Ruiz / Applied Surface Science 252 (2006) 7849–7855 7851

Fig. 1. XRD patterns of: (a) CP-FeSb, (b) SG-FeSb and (c) MC-FeSb.

Fig. 2. SEM micrographs of: (a) CP-Fe

It is also shown in Fig. 1 that the width of diffraction peaks is

in such an order: CP-FeSb > SG-FeSb > MC-FeSb, then the

particle size of the three samples will be in reverse order

according to Scherrer theory. This coincides with their specific

surface areas (38, 17 and 1.7 m2/g, respectively) and the SEM

results shown in Fig. 2. Raman analyses were done to find the

possible impurity Fe2O3 and Sb2O4 in CP-FeSb, SG-FeSb and

MC-FeSb, results are shown in Fig. 3. Sb2O4 exhibits five main

peaks at around 145, 199, 265, 403 and 465 cm�1, Fe2O3

exhibits five peaks around 226, 292, 411, 500 and 612 cm�1,

while FeSbO4 reveals almost no obvious peaks at 100–

700 cm�1 possibly because of its dark color. In accordance with

the XRD results, neither Fe2O3 nor Sb2O4 phase was observed

in these three samples.

The XPS responses in all the studied samples can be

assigned to elements Fe, Sb, O and the adventitious C, typical

spectra are shown in Fig. 4. Fig. 5 shows the Fe 2p and Sb 3d

XPS spectra of CP-FeSb, SG-FeSb and MC-FeSb. The Rs (i.e.

surface atomic ratio of Sb/Fe measured by XPS) and BE (i.e.

binding energy) of Fe 2p3/2, Sb 3d3/2 and O 1s are listed in

Table 1. The spectra of Fe 2p in Fig. 5 are typical Fe3+ signals

Sb, (b) SG-FeSb and (c) MC-FeSb.

Y. Huang, P. Ruiz / Applied Surface Science 252 (2006) 7849–78557852

Fig. 3. Raman spectra of: (a) a-Sb2O4; (b) MC-FeSb; (c) CP-FeSb; (d) SG-

FeSb and (e) Hematite Fe2O3.

Fig. 4. XPS spectra of single-phase FeSbO4 samples prepared by different

methods.

[20,21]. The BE of O 1s is around 530.3 eV, which is assigned

to O2�. The BE of Sb 3d3/2 bands is around 539.9 eV.

Table 1 indicates that Rs is always larger than corresponding

nominal atomic ratio of Sb/Fe (thereafter noted as Rn),

indicating that antimony segregation occurred. An increase in

Rn leads to an increase in Rs. Another interesting finding is that

Rs � 2 for the three single-phase FeSbO4 samples, and similar

results have been reported by Aso et al. [8–10] and Bowker

et al. [11,12]. According to the XPS results in Table 1, the

measured compositions of FeSbO4 surface layer on CP-FeSb,

SG-FeSb and MC-FeSb are FeSb1.8O5.6, FeSb2.1O6.7 and

FeSb2.0O6.4, respectively.

Fig. 5. Fe 2p and Sb 3d XPS spectra o

In ESCA analysis, the quantities of both Sb3+ and Sb5+ have

been summed together because the BE of Sb3+ and Sb5+ are too

close to allow individual quantifications [22]. However, the

XPS band of a mixture of Sb3+ and Sb5+ must be broader than

that of single Sb3+ or Sb5+, thus the variation of Sb5+/Sb3+ could

be monitored by the change in the full width at half maximum

(FWHM) of Sb 3d3/2. The maximum of FWHM would be

achieved when Sb5+/Sb3+ = 1, e.g. on the surface of pure Sb2O4.

Obviously, the surface concentration of Sb5+ against Sb3+ will

increase by the introduction of iron because of the formation of

FeSb(V)O4. The higher the content of iron, the larger ratio of

Sb5+/Sb3+, and consequently the lower FWHM of Sb 3d3/2. This

has been clearly confirmed by results in Fig. 6 (left), i.e. the

FWHM decreases with the increasing content of iron or the

decreasing nominal atomic ratio of Sb/Fe (thereafter noted as

Rn). No matter the surface antimony for Rn = 0.5 is single Sb5+

f CP-FeSb, SG-FeSb and MC-FeSb.

Y. Huang, P. Ruiz / Applied Surface Science 252 (2006) 7849–7855 7853

Fig. 6. Full width at half maximum (FWHM) of Sb 3d3/2 as a function of Rn and Rs (i.e. nominal and surface atomic ratios of Sb/Fe).

or a mixture of Sb5+ and Sb3+, the surface antimony for Rn = 1

and 2 is definitely a mixture of Sb5+ and Sb3+ because of its

larger FWHM.

Fig. 6 (right) indicates that an increase in Rs leads to an

increase in FWHM, and it can be concluded that the increase

in antimony segregation leads to an increase in surface

concentration of Sb3+. This agrees well with the results reported

by Zenkovets et al. [23].

4. Discussion

In the samples of single-phase FeSbO4 (proved by XRD and

Raman), there is an interesting phenomenon that the surface

Sb/Fe atomic ratio Rs, measured by XPS, is always close to 2,

despite those samples being prepared by quite different

techniques (e.g. mechanical method, 1.2 m2/g; slurry impreg-

nation [12], 5.9 m2/g; sol–gel, 17 m2/g; coprecipitation,

38 m2/g).

In literature, the surface layer of FeSbO4 grains has been

reported roughly as: (1) a layer of FeSbxOy (x > 1) species,

e.g. Fe(II)Sb(V)2O6 [7–10]; (2) a layer of solid solution of

Sb2O4–FeSbO4 [13,14]; (3) a thin ‘‘skin’’ [15] or epitaxy [16]

of Sb2O4 over FeSbO4. Apparently, the proposal of

‘‘FeSb2O6’’ well reflects the fact that the surface Sb/Fe

atomic ratios Rs of single-phase FeSbO4 are always close to 2.

In addition, the measured surface compositions of the three

single-phase samples of FeSbO4 are FeSb1.8O5.6, FeSb2.1O6.7

and FeSb2.0O6.4, all of which are very close to ‘‘FeSb2O6’’.

FeSb2O6 (JCPDS 7-349) has been included in the database

of International Center for Diffraction Data JCPDS-ICDD

1999, which crystalline structure was reported by Mason

and Vitaliano [24]. However, ‘‘FeSb2O6’’ should not be

‘‘Fe(II)Sb(V)2O6’’ as proposed in literature. The reason of

regarding FeSb2O6 as ‘‘Fe(II)Sb(V)2O6’’ might be an analogy

with Co(II)Sb(V)2O6, Ni(II)Sb(V)2O6, Cu(II)Sb(V)2O6,

Zn(II)Sb(V)2O6, etc. At first, most of the FeSbO4 samples

are prepared under oxidizing atmosphere, e.g. air, the ferrous

iron Fe2+ might be unstable. Secondly, Fe2+ was not observed

in our XPS study, nor in some other studies [12,13]. Thirdly,

unlike Cu2+, Zn2+, Ni2+, Co2+, etc., Fe2+ is strongly reductive,

and it can react with Sb5+:

2Fe2þ þ Sb5þ ! 2Fe3þ þ Sb3þ:

The coexistence of Sb5+ and Fe2+ in one compound is not

possible. Thus, the iron is ferric Fe3+, and the antimony is in

forms of Sb5+ and Sb3+. In addition, the FWHM studies in Fig. 6

confirmed that the antimony in the surface layer of single-phase

FeSbO4 is a mixture of both Sb5+ and Sb3+. Therefore, FeSb2O6

can be described as Fe(III)Sb(V)jSb(III)2�jO6. To balance the

valence, j = 1.5 and Sb5+/Sb3+ = 3. Fe(III)Sb(V)1.5Sb(III)0.5O6

may be also formulated as (FeSbO4)(Sb2O4)0.5. Although the

XPS results are well consistent with the suggestion about the

existence of stoichiometric compound of Fe2Sb2O6 on FeSbO4

surface, direct experimental evidence is still necessary. To be on

the safe side, the surface layer of FeSbO4 is described as

(FeSbO4)(Sb2O4)d, d � 0.5.

Besides of FeSbO4 and FeSb2O6, two other types of iron

antimonate have been reported: Fe(II)Sb(III)2O4 [25] and

Fe(II)2Sb(V)2O7. The former can be easily oxidized during

calcination in the air [26], it cannot be the surface species of

fresh FeSbO4 samples. ‘‘Fe2Sb2O7’’, which was ever erro-

neously suggested by Hussak and Prior to describe the

tripuhyite FeSbO4, does not exist [27,28]. This coincides with

the inference that the strongly reductive Fe2+ and the oxidative

Sb5+ should not coexist in one compound. In the database of

JCPDS-ICDD 1999, the file of Fe2Sb2O7 (JCPDS code 7-65)

has been deleted.

Fe–Sb mixed oxides with Sb/Fe > 1, such as CP-FeSb2 and

MC-FeSb2, are mixtures of Sb2O4 and FeSbO4. The presence of

excess Sb2O4 will increase the antimony concentration in the

surface layer of FeSbO4 component, and the composition of

this surface layer would be described as (FeSbO4)(Sb2O4)d,

d � 0.5.

Fe–Sb mixed oxides with Sb/Fe < 1 such as CP-Fe2Sb and

MC-Fe2Sb are mixtures of FeSbO4 and Fe2O3. For convenience

of discussion, the Sb/Fe ratio in the surface layer of Fe2O3

grains is noted as R0s, and that of FeSbO4 grains is noted as

Y. Huang, P. Ruiz / Applied Surface Science 252 (2006) 7849–78557854

Fig. 7. Illustrations of surface species of Fe–Sb mixed oxides with Sb/Fe < 1

(mol/mol).

R00s . At first, if R0s ¼ 0, R00s has to be very high because Rs are

found to be>1 on these two samples (cf. Table 1). However, the

existence of excess Fe2O3 should decrease R00s . Due to this

contradiction, it can be concluded that R0s 6¼ 0, i.e. antimony has

spread onto the surface of Fe2O3 grains. This is no surprise

according to the procedure of sample preparation. When a

mixture of Fe2O3 and Sb2O4 is calcined at high temperature,

Sb2O4 can spontaneously spread onto Fe2O3, such a behavior of

Sb2O4 is called ‘‘thermal-spreading’’ or ‘‘solid–solid wetting’’,

which is the phenomenon that a solid can spread onto the

surface of another solid simply by a calcination of their

mechanical mixtures [29–34]. Our previous work studied in

detail the thermal spreading of Sb2O3 and Sb2O4 over Fe2O3

surface [18,19,35]. As a result of the thermal spreading of

Sb2O4, the formation of FeSbO4 will preferably start on Fe2O3

surface.

2Fe2O3þ 2Sb2O4þO2 ! 4FeSbO4 (1)

Since the formation of FeSbxOy species such as FeSb2O6 is

highly possible, intermediate reactions could not be ruled out:

2FeSbO4þ Sb2O4 ! 2FeSb2O6 (2)

4FeSb2O6þ 2Fe2O3þO2 ! 8FeSbO4 (3)

These reactions will continue till Sb2O4 grains disappear,

and the remaining Fe2O3 will be encapsulated by a layer of

FeSbO4 as illustrated in Fig. 7. This is quite similar to the

mechanism of solid-reaction between Fe2O3 and MoO3

[36]. In a summary, the surface layer of FeSbO4 grains could

be described as (FeSbO4)(Sb2O4)d, d � 0.5, and the surface

of Fe2O3 grains would be a layer of FeSbO4.

The conventional Fe–Sb catalysts for selective oxidation

and ammoxidation of light hydrocarbons are mixtures of

FeSbO4 and a large amount of Sb2O4 [17], and the excess

Sb2O4 in the Fe–Sb catalysts is well-accepted to be a

necessity for better selectivity. Since pure Fe2O3 is a non-

selective catalyst, and the presence of excess Sb2O4 certainly

favors the elimination of Fe2O3 in catalysts, this seems to be

an easy explanation [37]. As found in this work, even if a

very small amount of Fe2O3 has not been completely

transformed during solid reactions, it will not be in the form

of free Fe2O3 grains but be encapsulated by a layer of

FeSbO4 (cf. Fig. 7).

5. Conclusion

The antimony in the surface layer of single-phase FeSbO4 is a

mixture of both Sb5+ and Sb3+ rather than single Sb5+, and the

surface composition is close to FeSb2O6, which could be

described as (FeSbO4)(Sb2O4)d, d = 0.5, and it is not

‘‘Fe(II)Sb(V)2O6’’ as suggested in literature. Fe–Sb mixed

oxides with Sb/Fe > 1 (mol/mol) are mixtures of FeSbO4 and

Sb2O4, but the surface of FeSbO4 grains would be a layer of

(FeSbO4)(Sb2O4)d, d � 0.5. Fe–Sb mixed oxides with Sb/Fe < 1

are mixtures of FeSbO4 and Fe2O3, and the surface of FeSbO4

grains would be a layer of (FeSbO4)(Sb2O4)d, d � 0.5, but the

remaining Fe2O3 would be encapsulated by a layer of FeSbO4.

Acknowledgements

The authors thank Dr. M. Genet, Unite de Catalyse et Chimie

des Materiaux Divises, Universite Catholique de Louvain,

Belgium, for assistance in XPS experiments. We would also

like to thank Dr. V. Cortes Corberan, C.S.I.C., Instituto de

Catalisis y Petroleoquımica, Spain, for his helpful discussions,

and his coworker Dr. R.X. Valenzuela for carrying out Raman

analysis. Financial support by Natural Science Foundation of

Jiangsu Province, China under Grant No. BK2005113 is

gratefully acknowledged.

References

[1] V. Fattore, Z. Fuhran, G. Manara, B. Notari, J. Catal. 37 (1975) 223.

[2] Y. Sasaki, Shokubai 40 (4) (1998) 257.

[3] S. Carrazan, L. Cadus, P. Dieu, P. Ruiz, B. Delmon, Catal. Today 32 (1996)

311.

[4] E. van Steen, M. Schnobel, R. Walsh, T. Riedel, Appl. Catal. 165 (1997)

349.

[5] Z. Magagula, E. Van Steen, Catal. Today 49 (1999) 155.

[6] L. Weng, B. Delmon, Appl. Catal. 81 (1992) 141.

[7] F. Sala, F. Trifiro, J. Catal. 41 (1976) 1.

[8] I. Aso, S. Furukawa, N. Yamazoe, T. Seiyama, J. Catal. 64 (1980) 29.

[9] I. Aso, T. Amamoto, N. Yamazoe, T. Seiyama, Chem. Lett. 4 (1980) 365.

[10] N. Yamazoe, I. Aso, T. Amamoto, T. Seiyama, Stud. Surf. Sci. Catal. 7

(1981) 1239.

[11] M. Bowker, C. Bicknell, P. Kervin, Appl. Catal. 136 (1996) 205.

[12] M.D. Allen, S. Poulston, E. Bithell, M.J. Goringe, M. Bowker, J. Catal.

163 (1996) 204.

[13] M. Carbucicchio, G. Centi, F. Trifiro, J. Catal. 91 (1985) 85.

[14] Z. Dziewiecki, A. Makowski, Reac. Kinet. Catal. Lett. 1 (1980) 51.

[15] M.D. Allen, M. Bowker, Catal. Lett. 33 (1995) 269.

[16] R. Teller, J. Brazdil, R. Grasselli, J. Chem. Soc. Faraday Trans. I 81 (1985)

1693.

[17] J.L. Callahan, B. Gertisser, US Patent 3197419, 1965.;

J.L. Callahan, B. Gertisser, US Patent 3338952, 1967.

[18] Y. Huang, G. Wang, R. Valenzuela, V.C. Corberan, Appl. Surf. Sci. 210

(2003) 346.

[19] Y. Huang, G. Wang, V.C. Corberan, Surf. Sci. 547 (2003) 55.

[20] E. Paparazzo, J. Electron Spectrosc. Relat. Phenom. 43 (1987) 97.

[21] K.J. Kim, D.W. Moon, S.K. Lee, K.H. Jung, Thin Solid Films 360 (2000)

118.

[22] R. Delobel, H. Baussart, J. Leroy, J. Grimblot, L. Gengembre, J. Chem.

Soc. Faraday Trans. I 79 (1983) 879.

[23] G.A. Zenkovets, G.N. Kryukova, S.V. Tsybulya, Kinet. Catal. 43 (2002)

731.

[24] B. Mason, C.J. Vitaliano, Min. Mag. 30 (1953) 100.

Y. Huang, P. Ruiz / Applied Surface Science 252 (2006) 7849–7855 7855

[25] B. Benaichouba, P. Bussiere, J. Friedt, J. Sancchez, Appl. Catal. 8 (1983)

237.

[26] E. Koyama, I. Nakai, K. Nagashima, Nippon Kagaku Kaishi 6 (1979) 793.

[27] P. Berlepschi, T. Armbruster, J. Brugger, A.J. Criddles, S. Graeser,

Mineral. Mag. 67 (2003) 31.

[28] E. Hussak, G.T. Prior, Min. Mag. 11 (1987) 302.

[29] Y. Xie, Y. Tang, Adv. Catal. 37 (1990) 1.

[30] Y. Xie, Y. Zhu, B. Zhao, Y. Tang, Stud. Surf. Sci. Catal. 118 (1998) 441.

[31] B. Pillep, P. Behrens, U. Schubert, J. Spengler, H. Knozinger, J. Phys.

Chem. B 103 (1999) 9595.

[32] F. Jentoft, H. Schmelz, H. Knozinger, Appl. Catal. 161 (1997) 167.

[33] H. Knozinger, E. Taglauer, Catalysis, vol. 10, The Royal Society of

Chemistry, Cambridge, 1993, p. 1.

[34] J. Leyrer, R. Margraf, E. Taglauer, H. Knozinger, Surf. Sci. 201 (1988)

603.

[35] Y. Huang, P. Ruiz, Antimony dispersion and phase evolution in Sb2O3–

Fe2O3 system, J. Phys. Chem. B, in press.

[36] F. Xu, Y. Hu, L. Dong, Y. Chen, Chin. Sci. Bull. 44 (1999) 14.

[37] V.P. Shchkim, G.K. Boreskov, S.A. Ven’yaminov, D.V. Tarasova, Kinet.

Katal. 11 (1970) 153.